← BACK
USER MANUAL

AEON — AI-Enhanced Order Navigator

User Manual  ·  Trade OS v0.4  ·  Last updated March 2026

Overview

AEON is a multi-account futures trading decision engine that combines real-time market analysis, AI-powered trade evaluation, and automated order execution into a single unified platform. It is designed for traders who want institutional-grade technical analysis applied consistently across multiple broker accounts — from manual decision support all the way to fully automated trade execution.

At its core, AEON uses Claude AI to evaluate market conditions across multiple timeframes. Every trade decision is backed by a rich context package: live OHLCV bars, EMA stacks, VWAP, key price levels (PDH/PDL, opening ranges, swing highs/lows), pattern detection, macro regime data, news sentiment, economic calendar events, and your account's current risk state. The AI synthesizes all of this into a directional verdict (LONG or SHORT) with a confidence score, which AEON uses to decide whether and how to enter a trade.

What AEON is not: AEON is not a set-and-forget black box. Even in Full Auto mode, you remain in control via risk limits, trade windows, daily caps, and kill switches. Understanding each module helps you configure AEON for your prop firm rules and personal risk tolerance.

Key Capabilities

  • Multi-account execution — fire bracket orders simultaneously across up to 4 prop firm accounts
  • AI position sizing — Conservative / Moderate / Aggressive strategies with market-quality multipliers
  • Multi-timeframe analysis — 1m, 5m, 15m, 1h, 4h, Daily regimes with EMA, RSI, ADX, MACD, ATR, VWAP
  • Full automation — scanner watches multiple tickers, AI decides, orders fire without manual intervention
  • Rapid Fire mode — high-frequency momentum scanner with per-ticker cooldowns and ATR-based brackets
  • Risk guardrails — daily loss limits, news blackouts, trade windows, and anti-hedge protection built in

System Architecture

AEON is split into two layers that communicate over a REST API:

Frontend (UI)

A single-page application served by nginx. All views live in app.js and are rendered client-side via hash-based routing (#/accounts, #/fullauto, etc.). The UI connects to the backend API on port 8000 for all data and actions.

Backend API (Flask)

A Python Flask application running in the aeon-api Docker container. It fetches live market data from TradingView (via tvDatafeed), calculates indicators, calls the Claude AI API, manages trade registry, executes orders through broker APIs, and persists state to SQLite.

Data Flow — Full Auto Trade

Scanner tick
Fetch bars (TV)
Calc indicators
AI analysis
Confidence gate
Risk checks
Fire bracket
Registry + log

Getting Started

Follow these steps when opening AEON for a new trading session:

  • 1. Check API status — the bottom of the sidebar shows a pulsing dot and API status. It should read online.
  • 2. Connect your accounts — go to Connections → Prop Firms and verify your broker API credentials are saved and active.
  • 3. Review risk settings — go to Risk Management and confirm daily loss limits are set for each account before any automation runs.
  • 4. Select accounts — on the Accounts page, check the accounts you want active for the session. This selection persists between page loads.
  • 5. Choose your mode — use the Decision Engine for manual/assisted trading, Full Auto for hands-off automation, or Rapid Fire for momentum scanning.

Accounts

The Accounts page is the session dashboard. It shows all connected trading accounts with their current balance, open P&L, daily P&L, and drawdown status. Accounts are toggled on/off using checkboxes — only selected accounts receive orders when you execute from the Decision Engine or automated modes.

  • Open PnL — pulled live from the broker's cash balance snapshot via an async poller; updates in real time without blocking the UI
  • Daily PnL — calculated from the session start balance vs current
  • Account selection — persisted automatically; survives page refreshes
  • Drawdown health dots — colored indicator dots on each account selector button show at a glance how close an account is to its daily loss limit (green → yellow → red)
  • Per-position flatten — individual open positions can be closed one at a time directly from the Accounts view, in addition to the global FLATTEN action
Prop Firm Tip: Always confirm the daily loss limit in Risk Management matches your prop firm's rules before enabling any automation. AEON enforces its own internal limit but cannot override the firm's platform limits.

AEON Trade Log

A chronological record of every trade AEON has opened or managed. Each entry shows the symbol, direction, entry price, stop loss, take profit, result (if closed), and the AI confidence score that triggered the trade. Use this to review AI decision quality over time.

  • Auto-sync on load — the trade log is refreshed automatically each time the page loads, so you always see the latest state without manually polling
  • Attribution — each entry is tagged as AEON (fired by automation) or MANUAL (entered by you), making it easy to separate automated performance from manual overrides

Audit Log

A low-level event log capturing every significant system action: scanner ticks, AI calls, order submissions, risk checks, errors, and manual overrides. Useful for debugging why a trade did or did not fire.

Tapout Status

The Tapout status bar is a persistent header element that shows each account's daily drawdown as a live percentage readout. It gives you an at-a-glance view of how much of your daily loss allowance has been consumed across all active accounts without navigating to the Accounts page.

  • Displays current daily loss as % of daily limit per account
  • Color shifts from neutral → amber → red as the limit is approached
  • When an account hits its limit, the tapout indicator locks red and automation halts for that account
DST-aware reset: Daily limits and tapout counters reset at 14:00 Pacific time year-round (automatically adjusts for daylight saving). This aligns with the futures market daily reset regardless of UTC offset.

Decision Engine

The Decision Engine is AEON's manual trading hub. It provides on-demand AI analysis for any supported symbol and presents three position sizing options. You review the AI's recommendation and fire the order yourself.

Workflow

  • Symbol selector — pick the instrument you want to analyze
  • Analyze — triggers a full multi-timeframe AI analysis; results display in the panel with direction, confidence, and reasoning
  • Position sizing — three strategies are presented after analysis:
StrategyRisk per TradeTypical Use
CONSERVATIVE0.5% of accountUncertain conditions, trending accounts
MODERATE1.5% of accountStandard daily trading
AGGRESSIVE3.0% of accountHigh-confidence setups only
  • ARM → EXECUTE — two-step confirmation to fire a bracket order (entry + stop + target) to all selected accounts
  • FLATTEN ALL — cancels all pending orders and closes all open positions on selected accounts immediately
  • Per-position flatten — individual positions can also be closed one at a time from the position list without flattening everything
  • HIGH risk alert — if the AI flags risk as HIGH, you must explicitly confirm before executing

Charts

The Decision Engine includes live TradingView charts for 1m, 5m, and 15m timeframes using real market data. Charts update on each analysis and are rendered using LightweightCharts.

Copy Trader

Copy Trader mirrors trades from a designated lead account to one or more follower accounts. When the lead account opens or closes a position, AEON replicates the action on all active follower accounts, optionally scaling size based on account balance ratios.

Direct Execute

Direct Execute allows you to send a specific bracket order to selected accounts without AI analysis. You manually specify symbol, direction, size, stop distance, and target distance. Useful for executing your own setups through AEON's multi-account order layer.

Gated Signals

Gated Signals receives external trade signals (e.g., from TradingView alerts or a webhook) and passes them through AEON's AI validation layer before executing. A signal is only acted on if the AI analysis agrees with the signal direction and meets the confidence threshold.

Use case: You have a TradingView strategy that fires alerts. Instead of blindly forwarding them to your broker, AEON validates each alert against live multi-timeframe AI analysis before deciding to execute.

Webhook Inbound

AEON exposes a dedicated inbound webhook URL that TradingView (or any external system) can POST alerts to directly. Each account gets its own unique inbound URL displayed in the Connections panel.

Payload Format

Webhooks support numeric position sizing — you can pass an exact contract count rather than relying on AEON's automatic sizing:

Example webhook payload

{"symbol":"MES","action":"LONG","qty":2,"stop":5,"target":10}

  • symbol — instrument to trade (e.g., MES, MNQ, MGC)
  • action — LONG or SHORT
  • qty — number of contracts (optional; omit to use AI sizing)
  • stop / target — stop and target distance in ticks (optional; omit to use ATR-based defaults)
Note: Inbound webhooks bypass AI analysis and confidence gating. Risk management limits (daily loss, trade window, anti-hedge) still apply.

Auto Signals

Auto Signals runs a continuous scanner across your configured ticker list. When the AI produces a qualifying signal on any ticker, AEON automatically fires the bracket order without requiring manual confirmation. Unlike Full Auto, Auto Signals respects a global confidence threshold and applies the full technical filter suite before executing.

Full Auto

Full Auto is AEON's complete end-to-end automation mode. The scanner continuously monitors all configured tickers, and for each one that passes its trade window and interval checks, the AI performs a full multi-timeframe analysis. Trades fire automatically when the AI returns LONG or SHORT above the confidence threshold.

Full Auto Scan Pipeline

  • 1. Trade window check — is the current time within the ticker's allowed trading hours?
  • 2. News blackout check — any high-impact economic event within ±15 minutes hard-blocks all trades for that scan cycle
  • 3. Technical filters — ADX trend strength, spread, session checks (skipped if filter disabled per-ticker)
  • 4. Scan interval — has enough time passed since the last AI call for this ticker? (minimum 5 min when filters are passing)
  • 5. Daily trade cap — has this ticker hit its max trades for the day?
  • 6. AI analysis — full multi-timeframe context sent to Claude
  • 7. Confidence gate — verdict must meet minimum confidence threshold
  • 8. Cooldown — minimum time between trades on this ticker
  • 9. Max concurrent positions — global cap on simultaneous open trades
  • 10. Anti-hedge — blocks opening a position opposite to an existing one on the same symbol
  • 11. Fire bracket — ATR-based stop loss and take profit, logged to registry
Daily loss halt: If an account's daily loss limit is reached, Full Auto immediately stops firing new trades for that account for the rest of the session, even if the scanner continues running.

Force Entry

Each ticker card in Full Auto has a FORCE button. This bypasses scan interval, technical filters, and confidence gate — the AI still runs, but any LONG/SHORT result fires immediately. Requires a two-step arm/confirm interaction to prevent accidental triggers.

Rapid Fire

Rapid Fire is a high-frequency momentum scanner designed to catch fast-moving setups. It runs on a much tighter loop than Full Auto and is configured per-ticker with its own ATR stop/target multipliers and cooldown settings.

Rapid Fire vs Full Auto

FeatureFull AutoRapid Fire
Technical filtersAppliedSkipped
Confidence gateAppliedSkipped (fires on any LONG/SHORT)
Global trade capAppliedSkipped
Max concurrent positionsGlobal cap appliedPer-ticker cap only (global skipped)
Anti-hedgeAppliedAlways applied
Per-ticker cooldownConfigurableConfigurable (default: off)
Daily loss haltAppliedApplied
Scan intervalPer-tickerPer-ticker (0 = random 10–30 min)
NEUTRAL verdictPossibleSkipped — always LONG or SHORT

Smart Account Independence

Rapid Fire executes on a per-ticker, per-account basis. Each account is evaluated and traded independently — one account hitting its daily cap or loss limit does not block Rapid Fire from continuing on other active accounts. This prevents a single account's state from stalling the entire scanner.

Caution: Rapid Fire is intentionally aggressive. Because it skips most safety gates, it can trade frequently. Always set per-ticker daily trade caps and ensure your daily loss limit is configured before enabling Rapid Fire.

Risk Management

Risk Management is the safety layer that governs all automated and manual trading activity. Settings here apply across Full Auto, Rapid Fire, and Auto Signals.

SettingDescription
Daily Loss Limit ($)Maximum dollar loss allowed per day per account. Automation halts when reached.
Max Concurrent PositionsGlobal cap on simultaneously open positions across all tickers (Full Auto only).
Per-Ticker Daily CapMaximum number of trades allowed per ticker per session.
Global Daily CapMaximum total trades across all tickers in a session (Full Auto).
Cooldown (seconds)Minimum wait between consecutive trades on the same ticker.
Trade WindowTime range (UTC) during which new trades may be opened for a ticker.
News BlackoutAutomatic block on new trades within ±15 minutes of a high-impact economic event.
Futures Market CutoffConfigurable per-ticker schedule to stop new orders before CME settlement or close. Prevents orders from being sent into low-liquidity market-close windows.
Risk Config ToggleMaster on/off switch that enables or disables all risk enforcement globally. Turn off only for testing; leave on in live trading.
When daily_loss_halted is triggered, a banner appears in the UI and all automation stops firing for that account. It resets at 14:00 Pacific time daily (DST-aware — always aligns with the futures market rollover regardless of time of year).

Model Config

The Model Config panel lets you choose which Claude AI model AEON uses for trade analysis and control inference parameters without editing config files.

Each function (Decision Engine, Full Auto scanner, Rapid Fire scanner, Signal Gate, Exit Analysis) has its own independently configured model selection.

ModelSpeedBest For
claude-haiku-4-5Fastest / CheapestRapid Fire, high-frequency scanning
claude-sonnet-4-6Balanced (Default)Full Auto, Decision Engine
claude-opus-4-6Most Capable / SlowestHigh-stakes manual analysis
Note: Temperature and max token settings are optimized per-function and are not exposed as UI controls. Use the model selector to trade off speed vs quality.

Connections

The Connections section manages all external integrations:

  • Prop Firms — broker API credentials (Tradovate, NinjaTrader, Apex, TopstepFX). Credentials are stored server-side and never exposed in the UI after saving.
  • TradingView — tvDatafeed credentials used to fetch live market data bars. AEON requires a TradingView account for chart data.
  • Discord — optional webhook for trade notifications and alerts to a Discord channel.
  • Databento — planned future integration for professional CME tick data. Currently a placeholder; TradingView remains the active data source.

Supported Symbols

SymbolInstrumentType
GCGold (Full-size)Metals
MGCGold (Micro)Metals
SISilver (Full-size)Metals
SILSilver (Micro)Metals
MSISilver (Micro — smaller)Metals
ESS&P 500 E-MiniEquity Index
MESS&P 500 MicroEquity Index
NQNasdaq-100 E-MiniEquity Index
MNQNasdaq-100 MicroEquity Index
CLCrude Oil (Full-size)Energy
MCLCrude Oil (Micro)Energy

How AI Analysis Works

Every trade decision — whether triggered manually from the Decision Engine or automatically by the scanner — goes through the same AI analysis pipeline. AEON builds a comprehensive market context package and sends it to Claude AI, which returns a directional verdict and confidence score.

The AI does not guess. It is given structured, quantified data about every relevant market condition and instructed to reason through the setup systematically before producing a verdict. The context package includes 14 distinct data categories described below.

Indicators & Data Passed to AI

1 — OHLCV Bars

200 bars minimum across 2–3 timeframes (specific timeframes depend on the selected trade style). Recent candles are shown in structured format with direction markers.

2 — EMA Stack (9 / 21 / 50 / 200)

All four EMAs are calculated per timeframe. The AI receives raw values, alignment classification (BULLISH STACK, BEARISH STACK, MIXED), and price position relative to EMA9/EMA21.

3 — VWAP (Session)

Typical-price VWAP calculated from session start (23:00 UTC). The AI is told whether price is above or below VWAP and the distance in ATR units.

4 — Key Price Levels

  • PDH / PDL — Prior Day High / Low
  • ONH / ONL — Overnight High / Low
  • CSH / CSL — Current Session High / Low
  • Opening Ranges — ASIA (23:00), LONDON (07:00), NY (13:30) first-15-min ranges
  • Swing Highs / Lows — Up to 4 most recent pivots from the anchor timeframe, deduped within 0.1%

All levels are presented with ATR-normalized distance so the AI can judge proximity in volatility-adjusted terms.

5 — Market Structure

HH/HL/LH/LL pattern classification (BULLISH STRONG, BEARISH STRONG, BULLISH MIXED, BEARISH MIXED, RANGING, UNKNOWN) derived from comparing the last two swing highs and lows.

6 — Momentum Indicators

  • RSI(14) — with overbought/oversold labels
  • ADX(14) — TRENDING vs RANGING/WEAK classification
  • MACD(12,26,9) — line, signal, histogram with momentum label
  • ATR(14) — used for normalizing all distance calculations
  • Squeeze State — Bollinger Bands vs Keltner Channel: SQUEEZE / FIRING / RELAXED
  • MACD Acceleration — histogram slope over last 3 bars: ACCELERATING / DECELERATING / FLAT
  • Volume Delta — 20-bar signed volume: STRONG_BUY / BUY / NEUTRAL / SELL / STRONG_SELL

7 — Candlestick & Price Action Patterns

Pattern engine detects: Bullish/Bearish Engulfing, Fair Value Gaps (FVG), Break of Structure (BOS), and Liquidity Sweeps. Up to 3 FVGs per timeframe included.

8 — Macro Regime

VIX level and label, NQ (Nasdaq-100) trend and RSI(14), and overall macro regime classification. For metals/energy symbols, a note is included that NQ is context-only, not a direct driver.

9 — Economic Calendar

High-impact events within the next 60 minutes are listed with exact distances. If an event is within ±15 minutes, a hard news blackout is enforced and the AI is not called — no trades fire.

10 — News Sentiment

Last 48 hours of headlines from Google News RSS for the symbol. Bullish/bearish keyword counts produce an overall sentiment label: BULLISH, BEARISH, MIXED, or NEUTRAL.

11 — Account Risk State

Current daily loss consumed ($ and %), headroom remaining, and alert level. If the account is at critical risk, the AI factors this into its sizing and confidence recommendations.

12 — Recent Trade Outcomes

Last 5 closed AEON trades for the symbol: direction, entry/exit prices, P&L, confidence score at entry, and setup type. Helps the AI avoid repeating patterns that have been failing recently.

13 — Setup Win Rates

Historical win rates broken down by setup type (Pullback, Breakout, FVG Fill, etc.) are included so the AI can calibrate its confidence scoring against real performance data.

14 — Liquidity Map

A synthesized map of where resting liquidity is likely to be located above and below the current price. Derived from clustered swing highs/lows, equal highs/lows, and prior session extremes. The AI uses this to assess whether price is hunting liquidity, if a stop run is likely before a reversal, and where the next magnet level sits.

Confidence Scoring

The AI returns a confidence score from 35 to 100 for every verdict. AEON uses this score to gate trade execution.

RangeLabelAction (Full Auto)
75 – 100STRONGFires immediately
50 – 74MODERATEFires if above configured threshold
35 – 49LOWBlocked unless threshold is set low
NEUTRALNever fires (Full Auto skips NEUTRAL)

In Rapid Fire mode, NEUTRAL verdicts are suppressed — the AI always returns LONG or SHORT, and the confidence gate is skipped entirely. Any LONG/SHORT fires.

Scalp style cap: When trade style is set to Scalp, the AI operates under a hard confidence ceiling of 60. This is intentional — scalp setups are inherently lower conviction; the AI is instructed never to exceed 60 regardless of setup quality. Day Trade, Swing, and Position styles have the full 35–100 range.

Trade Style Modes

The timeframes and bar counts used for AI context change depending on the selected trade style:

StyleFastest TFPrimary TFAnchor TF
Scalp1m (200 bars)5m (200 bars)15m (100 bars)
Day Trade5m (200 bars)15m (200 bars)1h (100 bars)
Swing15m (200 bars)1h (100 bars)4h (80 bars)
Position1h (100 bars)4h (80 bars)Daily (60 bars)

The AI system prompt is also tuned per style — scalp prompts emphasize momentum and precision timing, while swing/position prompts weight structure and macro alignment more heavily.

Diagnostics

The Diagnostics section provides real-time visibility into the scanner internals and system health. Global controls are also accessible here.

  • Full Auto diagnostics — shows each scanner tick, which tickers were checked, what stage they reached in the pipeline, and why a trade did or did not fire
  • Rapid Fire diagnostics — same output for the Rapid Fire scanner loop
  • System Info — live CPU usage, memory consumption, and process stats from the backend container (via /api/sysinfo). Useful for spotting resource pressure during heavy scanning sessions.
  • Global controls — emergency stop, scanner restart, and session reset accessible from a single panel without navigating between sections

Use diagnostics to verify your configuration is working correctly before leaving AEON running unattended, and to investigate why expected trades did not execute.

Glossary

TermDefinition
ATRAverage True Range — a measure of market volatility used to size stops and targets
Bracket OrderA combined order set: entry + stop loss + take profit, submitted simultaneously
BOSBreak of Structure — price breaking a prior swing high or low with confirmation
CooldownMinimum wait time between consecutive trades on the same ticker
EMA StackThe alignment of EMA(9), EMA(21), EMA(50), EMA(200) relative to each other and price
FVGFair Value Gap — an imbalance in price where no trading occurred; often acts as a magnet
FlattenCancel all open orders and close all positions on selected accounts immediately
HH/HL/LH/LLHigher High / Higher Low / Lower High / Lower Low — market structure classification
MTFMulti-Timeframe — analysis across multiple chart timeframes simultaneously
News BlackoutAuto-block on new trades within ±15 min of a high-impact economic event
ONH / ONLOvernight High / Overnight Low — price extremes during the overnight session
PDH / PDLPrior Day High / Prior Day Low — key reference levels from the previous session
RegimeThe macro market state (e.g., TRENDING BULLISH, RANGING) derived from multi-TF analysis
SqueezeBollinger Bands inside Keltner Channel — signals compressed volatility before expansion
Trade WindowConfigured time range (UTC) during which a ticker is allowed to open new trades
VWAPVolume Weighted Average Price — institutional reference level from session open
TapoutDaily drawdown status indicator showing % of daily loss limit consumed per account
Liquidity MapSynthesized map of clustered resting orders (stop-loss pools) above and below current price
AttributionTag on each trade log entry indicating whether the trade was fired by AEON automation or entered manually
DST-aware resetDaily limit reset that automatically adjusts for Daylight Saving Time, always firing at 14:00 Pacific regardless of UTC offset
Per-position flattenClosing a single open position without affecting other open positions on the account
Smart account independenceRapid Fire's behavior of evaluating and trading each account separately so one account's state doesn't block others